Signal-to-Noise Ratio (SNR) in astrophotography is a difficult topic for beginners. As someone who didn’t have any exposure (no pun intended) to photography before starting astrophotography, I wasn’t initially remotely interested in learning more about this topic and just wanted to take “pretty images”. However, after about 6 months of doing astrophotography, I realized how much I wanted to investigate what SNR really is and what equipment upgrades would help take my images to the next level.
So, for those like me who want to learn more, this page first discusses at a high level what light/calibration frames are; next, I share how different factors impact SNR; then, I share an interactive SNR calculator spreadsheet that I put together which is an excellent hands-on tool to help quantify how those different factors impact SNR at a theoretical level; and finally, I share some additional resources that are also great at explaining this topic.
Light and Calibration Frames
When you see a final image someone took of a deep sky object, you aren’t seeing the result of a single photograph. Instead, you’re actually seeing the combination of likely hundreds of images of the actual object and images that helped calibrate those images of the actual objects. These are called “light” and “calibration” frames, respectively.
Light frames are the actual images taken of the deep sky object of interest. As we’ll see later when discussing the Signal-to-Noise Ratio (SNR) below, it is best to take a large number of these long exposures to help best bring out the faint light/details and to reduce the amount of noise from unwanted sources. This is the most straight forward type of image that is included in astrophotography.
Calibration frames are broken down into 3 types, with the third having two options depending on your set-up:
- Darks: camera sensors have an inherent dark current that contributes to noise during long exposures. These images are taken by keeping the same settings as your light frames, but covering the optics with the lens cap or telescope cover. In effect, you are simply looking to capture and measure the dark current (plus read noise) from your camera sensor with no other light getting to the sensor. (Note: some dedicated cooled astrophotography cameras do not have a significant dark current when compared to read noise nor have amp glow; if this is the case, then you may not need to use dark frames and instead bias frames will take its place in the calibration formula below. For example, I do not use dark frames with my ASI2600MC Air.)
- Flats: many lenses and telescopes will result in vignetting on the camera sensor, where the corners and edges of the light frames will appear darker than the middle. Additionally, dust spots are likely to appear somewhere within the imaging train, whether on the camera sensor, camera/telescope lens, or telescope mirror. Properly taken flats will eliminate these issues. To take them, I prefer to take a white t-shirt and use a rubber band to keep it in place and tight over the end of the lens/telescope. Keeping all other settings the same as lights (especially the focus), simply reduce the exposure time until you get the histogram about 1/3 from the left. (Note: if you are using a computer like an ASIAir, this can automatically calculate the exposure time needed for flats, which makes them very easy.)
- Bias (also “Dark Flats”): every single camera has a read noise component when any image is taken and saved/stored. This noise should be random and should not show as fixed pattern noise such as obvious lines in a stretched bias image. To take these, again keep other camera settings the same as lights and set the exposure time to as short as possible with your camera. This is done to remove all other sources of “light” to simply capture what that read noise output is. Some camera sensors may experience amp glow on the side of the sensor; in those cases, the bias frames will need to be taken at the exact same exposure length as the flat frames (and are thus referred to as “dark flats”). If you are unsure whether you need dark flats, it is highly recommended to use them over a true “bias” frame.
These calibration frames are used to “calibrate” the individual light frames using this formula: . I recommend not overthinking how many are “optimal” to take and that you take 30+ of them each.
Additionally, darks and biases (or “dark flats”) can be taken anytime without the need to be attached to a lens/telescope. These may also be reused for many months as a “dark/bias library”, though it should be noted that the “dark library” does need the camera sensor to be at the same temperature as the lights (which is one reason a cooled astrophotography camera is very useful). Flat frames, however, must be taken for each imaging session (and a separate set of flats with each filter, if using multiple filters during a session).
Finally, rather than trying to describe in text the exact methodology of how these calibration frames work, I highly recommend watching this video from Dr. Adam Block called “Calibration of Astronomical Images (PixInsight/WBPP)“. (Note: you can ignore that it says “PixInsight/WBPP”, as the information presented in this video is not relevant to just PixInsight or Weighted Batch Pre Processing (WBPP).)
Signal-to-Noise Ratio (SNR)
As mentioned before, a final astrophotography image is actually made from (likely) hundreds of individual images of the deep sky object and additional calibration frames. The reason for this is that most deep sky objects are very faint and stacking this many frames are needed to have a sufficient Signal-to-Noise Ratio (SNR). SNR is defined at a high level as:
The “signal” is defined as the actual light that you capture from the deep sky object, which is obviously what we want we are capturing an image of (like the photons coming from the Andromeda Galaxy).
The “noise” comes from two sources, “wanted sources” and “unwanted sources”. The object you’re taking images of will have some noise associated with it, especially in the form of “shot noise” (more on this at the end of this section below), so this ends up being noise from a “wanted source”; there is no way to remove this source of noise because we’re trying to capture these photons, and this is present even in images taken in space telescopes (like Hubble or James Webb) where this is no (significant) light pollution at all. However, all other forms of noise including dark current, read noise, and light pollution are from “unwanted sources” which we try to reduce as much as possible.
The detailed formula for SNR for a single exposure is:
The detailed formula for SNR for a stack of exposures is the same as above multiplied by the square root of the number of subexposures taken. To improve the SNR in a stack of images, there are two options: increase the signal or decrease the noise.
The main ways to increase the signal in an image stack (without changing equipment) are through longer subexposure times or by taking more subexposures. In both cases, the aggregate number of photons from the deep sky object reaching the camera sensor increases.
Reducing noise can be achieved in a few ways, including reducing dark current through camera cooling, reducing read noise through different camera settings (i.e., a higher ISO/gain setting typically reduces read noise at the consequence of lower dynamic range), or by reducing sky noise/light pollution with darker skies.
Finally, as the stacked image SNR is multiplied by the square root of the number of exposures, there is a diminishing return when taking more images. For example, going from 1 to 2 images will increase the stacked SNR by ~41% (i.e., √(2/1) = 1.414); however, if you already have 100 images, then adding 1 more image will only improve your stacked SNR by ~0.5% (i.e., √(101/100) = 1.00498).
Interactive SNR Calculator Spreadsheet
By now you may have an basic understanding of what lights/calibration frames are and how they are used to improve SNR. But it can be hard to know right away what factor is dominating your unwanted noise; perhaps instead of purchasing a new camera with better cooling to reduce the dark current, your location needs to change because sky glow/light pollution is dominant.
As I did my deep dive into the formulas for SNR and how the various factors impact it on a purely theoretical level, I created a Google Sheets spreadsheet (for One-Shot-Color imaging only) that takes the multiple adjustable inputs and calculates the SNR. For example, this can be used to see how much SNR changes if you imaged from a Bortle 4 location instead of your light polluted Bortle 7 home with the exact same equipment.
Additional calculations are also included on the main Analysis tab. There are a number of caveats and important notes added on the Home/Information tab. And since all data for deep sky objects and equipment were entered manually, I obviously could not include everything (so when you download your own copy you are encouraged to enter your own information, as needed).
Finally, those with the link are given “Commenter” permissions so that it can be downloaded either as a Google Sheet or Excel file. This is done so that multiple people do not try to access this at the same time and overwrite changes made to equipment or imaging sessions simultaneously, which would cause chaos/confusion when trying to see how a change to one factor impacts SNR.
If you see anything in that spreadsheet that is incorrect or unclear, feel free to leave a comment or directly email me (at jbaltusastro@gmail.com). Any other comments (especially inappropriate ones) will be promptly removed.
Additional Resources
For a greater and deeper understanding of SNR and more discussion for how various factors impact it, these resources were very helpful to me: